Machine learning has improved the accuracy of genetic risk prediction for type 1 diabetes, moving beyond limitations of existing scores. This breakthrough enables earlier identification of at-risk individuals.